Published September 20, 2019 | Version 1.0
Dataset Open

Layout Analysis Groundtruth for the RVL-CDIP Dataset

Creators

  • 1. omni:us

Contributors

  • 1. CVC

Description

The purpose of this dataset is to develop and evaluate layout analysis techniques. More specifically it is focused on the classification of indiviual words and the detection of semantic regions described by boxes.

Files

dataset.zip

Files (35.6 MB)

Name Size Download all
md5:94bfcad49ed27f5a2fe9f0e4286106a3
35.6 MB Preview Download
md5:6866093723236d065bfde400d078fcac
1.4 kB Preview Download

Additional details

References

  • Pau Riba et al.: "Table Detection in Invoice Documents by Graph Neural Networks", ICDAR, 2019.